Specification of prior distributions under model uncertainty
Specification of prior distributions under model uncertainty
We consider the specification of prior distributions for Bayesian model comparison, focusing on regression-type models. We propose a particular joint specification of the prior distribution across models so that sensitivity of posterior model probabilities to the dispersion of prior distributions for the parameters of individual models (Lindley's paradox) is diminished. We illustrate the behavior of inferential and predictive posterior quantities in linear and log-linear regressions under our proposed prior densities with a series of simulated and real data examples.
bayesian inference, bic, generalised linear models, lindley’s paradox, model averaging, regression models
Southampton Statistical Sciences Research Institute, University of Southampton
Dellaportas, Petros
df8947f6-37ea-4e68-8967-eb43f777a5fd
Forster, Jonathan J.
e3c534ad-fa69-42f5-b67b-11617bc84879
Ntzoufras, Ioannis
334f2302-b765-48d1-90b6-121ec8e31131
13 May 2009
Dellaportas, Petros
df8947f6-37ea-4e68-8967-eb43f777a5fd
Forster, Jonathan J.
e3c534ad-fa69-42f5-b67b-11617bc84879
Ntzoufras, Ioannis
334f2302-b765-48d1-90b6-121ec8e31131
Dellaportas, Petros, Forster, Jonathan J. and Ntzoufras, Ioannis
(2009)
Specification of prior distributions under model uncertainty
(S3RI Methodology Working Papers, M09/10)
Southampton, UK.
Southampton Statistical Sciences Research Institute, University of Southampton
22pp.
Record type:
Monograph
(Working Paper)
Abstract
We consider the specification of prior distributions for Bayesian model comparison, focusing on regression-type models. We propose a particular joint specification of the prior distribution across models so that sensitivity of posterior model probabilities to the dispersion of prior distributions for the parameters of individual models (Lindley's paradox) is diminished. We illustrate the behavior of inferential and predictive posterior quantities in linear and log-linear regressions under our proposed prior densities with a series of simulated and real data examples.
Text
s3ri-workingpaper-M09-10.pdf
- Other
More information
Published date: 13 May 2009
Keywords:
bayesian inference, bic, generalised linear models, lindley’s paradox, model averaging, regression models
Identifiers
Local EPrints ID: 66215
URI: http://eprints.soton.ac.uk/id/eprint/66215
PURE UUID: ff3544ee-336f-42a0-97c7-17574ca9065c
Catalogue record
Date deposited: 13 May 2009
Last modified: 14 Mar 2024 02:37
Export record
Contributors
Author:
Petros Dellaportas
Author:
Jonathan J. Forster
Author:
Ioannis Ntzoufras
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics